diff options
| author | obrien <obrien@FreeBSD.org> | 2004-04-22 21:37:46 +0000 |
|---|---|---|
| committer | obrien <obrien@FreeBSD.org> | 2004-04-22 21:37:46 +0000 |
| commit | cbc25a780d4495552e787db7a73df5de107da241 (patch) | |
| tree | 29e48d09f3985d9afe6fc0df40337e0e7230d20e /sys/kern | |
| parent | a0d6a22b415d781561c98c0a04fae9ad59350fe9 (diff) | |
| download | FreeBSD-src-cbc25a780d4495552e787db7a73df5de107da241.zip FreeBSD-src-cbc25a780d4495552e787db7a73df5de107da241.tar.gz | |
There was a thread on "unusually high load averages" when running under
sched_ule, in January 2004. Looking at this, "pagezero" is (one of) the
culprit(s). We had no provision for processes with P_NOLOAD set. With
pagezero not running at PRI_ITHD, kseq_load_{add,rem} count pagezero as
another-normal-process, thus the "expected-plus-one" load reported in
the above thread.
Submitted by: Nikos Ntarmos <ntarmos@ceid.upatras.gr>
Diffstat (limited to 'sys/kern')
| -rw-r--r-- | sys/kern/sched_ule.c | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/sys/kern/sched_ule.c b/sys/kern/sched_ule.c index a2268d3..a65fdf1 100644 --- a/sys/kern/sched_ule.c +++ b/sys/kern/sched_ule.c @@ -357,7 +357,7 @@ kseq_load_add(struct kseq *kseq, struct kse *ke) if (class == PRI_TIMESHARE) kseq->ksq_load_timeshare++; kseq->ksq_load++; - if (class != PRI_ITHD) + if (class != PRI_ITHD && (ke->ke_proc->p_flag & P_NOLOAD) == 0) #ifdef SMP kseq->ksq_group->ksg_load++; #else @@ -380,7 +380,7 @@ kseq_load_rem(struct kseq *kseq, struct kse *ke) class = PRI_BASE(ke->ke_ksegrp->kg_pri_class); if (class == PRI_TIMESHARE) kseq->ksq_load_timeshare--; - if (class != PRI_ITHD) + if (class != PRI_ITHD && (ke->ke_proc->p_flag & P_NOLOAD) == 0) #ifdef SMP kseq->ksq_group->ksg_load--; #else |
